pISSN 1899-5241
eISSN 1899-5772 2(56) 2020, 201–219
Tinashe Chuchu, Department of Marketing Management, Faculty of Economic and Management Sciences, University of
Pretoria, e-mail: tinashe.chuchu@up.ac.za, https://orcid.org/0000-0001-7325-8932
THE RESPONSE OF ZIMBABWE TOBACCO EXPORTS
TO REAL EXCHANGE RATES VOLATILITY: 1980–2019
Knowledge Mutodi
1, Eugine Tafadzwa Maziriri
2, Tinashe Chuchu
3 1University of Zimbabwe, Zimbabwe2University of the Free State, South Africa 3University of Pretoria, South Africa
Abstract. The focus of this study was on investigating the
response of tobacco exports to real exchange rates and real exchange rate volatility and other factors in Zimbabwe using secondary data spanning from 1980 to 2019. Bilateral nominal exchange rates and time-variant weights of Zimbabwe’s 10 major trading partners were calculated and used to compute the real exchange rate index. The time-dependent weighting system was used to better represent the evolution of trade pat-terns in the index. The arithmetic method was employed for computing the index. Generalized autoregressive conditional heteroskedasticity (GARCH) and autoregressive conditional heteroscedasticity (ARCH) models were used to generate the real exchange rate volatility index. The export response func-tion was adopted as the tobacco exports response model. The variables in the tobacco exports response model were the real-world Gross Domestic Product (GDP), real exchange rate, terms of trade, real exchange rate volatility and dollarization. A vector error correction model (VECM) was used to estimate the response of tobacco exports to real exchange rate, real ex-change rate volatility and other factors. The VECM results in-dicated that real world GDP was insignificant in both the short and long run. In the long run, the real exchange rate apprecia-tion had a negative impact on tobacco exports. Conversely, in the short run, the depreciation of real exchange rate had a positive impact on tobacco exports. Hence, the government has to adopt other mechanisms that reduce uncertain move-ments of exchange rates.
Keywords: real exchange rate (RER), real exchange rate
vol-atility (RERV), vector error correction model (VECM)
INTRODUCTION
Zimbabwe is a small open economy (Nyanga et al., 2020; Sithole, 1996). Its ratio of gross domestic product (GDP) to world GDP was 0.03 in 2014 and 0.02 in 2015 (World Bank, 2016). In 1960–2018, the average GDP per capita in Zimbabwe was USD 1220.33 (Trade Econom-ics, 2020). Furthermore, Zimbabwe reached an all-time highest increase in GDP of USD 1524.30 in 1998 and a record low of USD 725.60 in 2008 (Trade Econom-ics, 2020). However, in 2015, imports and exports as a percentage of GDP were 48.78% and 26.25%, respec-tively (ZimTrade, 2015). Zimbabwe’s major exports are minerals followed by agriculture. Among these major exports, the share of minerals in total exports in 2012 was 64% of which 26.9% were diamonds, 26.1% was gold and 0.7% were nickel and others. The share of agricultural exports in total exports in 2009 was 35%, of which 61% was tobacco, 22% was cotton, 11% was raw sugar and 3% were horticultural products. Tobac-co as the dominant cash and export crop resulted in USD 855 million of foreign exchange earnings in 2015 and contributed about 26% to agricultural GDP in 2009 (Government of Zimbabwe, 2016; Tobacco Industry…, 2015). The major imports are diesel fuel, unleaded gas, maize, wheat, crude soya bean oil and others. The share of fuels and cereals in 2014 imports was 25% and 5%, respectively (ZimTrade, 2015). However, agricultural
imports have recently surged as the share of food im-ports in merchandise imim-ports has increased from 14.26% in 2013 to 15.14% in 2015 (Buigut, 2015). Low agricul-tural food production has increased the appetite for food imports and resulted in a general economic dependence on imports.
Large depreciations and high volatility of the real exchange rate were connected to major structural shifts, namely the economic structural adjustment (ESAP) in 1992 and particularly the land reform in 2000. Under the dollarization policy, the real exchange rate appre-ciated against Zimbabwe’s major trading partners. This signaled a loss of external competitiveness of Zimba-bwean exports. The terms of trade were significant in both the short and long run. The real exchange rate vola-tility indicated that the impact of high fluctuations on to-bacco exports was negative in the short run and positive in the long run. Dollarization was significant in the short run and insignificant in the long run. Hence, apprecia-tion and random movements of the real exchange rate were found to affect the exports of tobacco, a dominant export commodity in agriculture. The study recom-mends the government to devalue the real exchange rate without changing the nominal exchange rate value for exports to remain competitive in both the domestic and international market. Also, the authorities can promote futures markets to hedge against uncertain movements of exchange rates. Terms of trade can be improved by raising taxes on importable goods. High interest rates can be set to attract high capital inflows to reduce tight liquidity conditions in the economy.
Over the past two decades, the economy has wit-nessed great changes in production volumes and general patterns of trade. This transformation has been a result of the introduction of several micro- and macroeconomic policy changes (World Bank, 2016). However, the intro-duction of these policies has seen major structural shifts in the economy. The major causes of structural shifts were the attainment of independence in 1980, the intro-duction of the economic structural adjustment program (ESAP) in 1992, the introduction of the land reform pro-gram in 2000 and dollarization of the economy in 2009 (United Nations…, 2009). Independence opened up the country to international trade with the abolition of in-ternational trade sanctions (Mumbengegwi, 2002). The introduction of ESAP in 1992 further opened the coun-try to international trade. The government implemented monetary, fiscal, trade, exchange rate and agricultural
policy reforms (Sithole 1996). The land reform, intro-duced around 2000 to address the uneven land distribu-tion between black and white farmers, quickly closed up the economy.
Problem statement
Maintaining an internal and external balance remains a major challenge in Zimbabwe (Reserve Bank of Zim-babwe, 2014). The internal and external imbalances be-fore 2009 have been reflected by a slow erratic growth in Gross Domestic Product, hyperinflation, foreign ex-change shortages and BOP deficits (World Bank, 2016). The dollarization introduced in 2009 was followed by fast, short-run economic growth, but later the country became besieged with liquidity challenges and wit-nessed sluggish growth rates (AFDB, 2015). Dollariza-tion led to liquidity challenges causing weak aggregate demand that affected economic growth. Fiscal space has greatly diminished due to high taxation levels, suc-cessive government budget deficits, and the increase in both domestic and external debt (Reserve Bank of Zimbabwe, 2016a). The domestic debt averaged 162% in 2013, and increased to an average of 202.4% of GDP in 2016 (Reserve Bank of Zimbabwe, 2016b). The debt build-up is due to money being borrowed to finance cur-rent account deficits, driving an increase in the debt-to-GDP ratio (AFDB, 2015). However, the introduction of bond coins and notes in 2014 and 2016, respectively, failed to address the liquidity challenges facing the economy. With the introduction of bond notes, the prob-lem quickly resurfaced as inflation and depreciation of bond notes threatening the economic growth (Reserve Bank of Zimbabwe, 2016a). Depreciation of the local currency was necessary to cure its overvaluation.
The overvalued Zimbabwean dollar, then the U.S. dollar, and now the bond notes are weakening the exter-nal sector competitiveness, resulting in severe interexter-nal and external imbalances or current account deficits (Re-serve Bank of Zimbabwe, 2015). This further discour-ages foreign direct investment. The presence of current account deficits and/or capital outflows weakens inter-nal liquidity inflows, leading to liquidity problems. The depreciation of bond notes against the U.S. dollar is fur-ther affecting the economy.
The internal and external imbalances have nega-tively affected incentives to produce goods for both the domestic and the export markets (Saungweme, 2012). Both domestic and foreign markets are greatly
depressed; this has resulted in reduced output and ex-ports in most sectors and increased imex-ports, especially of food. The results have been the mounting trade and current account deficits, sluggish economic growth, and higher unemployment rates. Further, increased govern-ment deficits and growing debt have constrained the government’s ability to service both domestic and in-ternational debt, sustain its real exchange rate policy and provide enough liquidity to oil the economy (World Bank, 2016). Hence, the broad objective of this study is to investigate the response of Zimbabwean tobacco exports to the real exchange rate and real exchange rate volatility in 1980–2019.
Table 1. Research objectives and questions
i. Compute Zimbabwe’s RER
index i. How has the Zimbabwean real exchange rate index evolved?
ii. Compute Zimbabwe’s RER
volatility index ii. How has the Zimbabwean RER index volatility evolved? iii. Estimate the response of
tobacco exports to import intensity (world demand) in both the short and long run
iii. How have tobacco exports responded to import intensity (world demand) in both the short and long run? iv. Estimate the response of
tobacco exports to the RER index in both the short and long run
iv. How have tobacco exports responded to the RER index in both the short and long run?
v. Estimate the response of to-bacco exports to terms of trade in both the short and long run
v. How have tobacco exports responded to terms of trade in both the short and long run? vi. Estimate the response of
tobacco exports to the RER volatility index in both the short and long run
vi. How have tobacco exports responded to the RER volatil-ity index in both the short and long run?
vii. Estimate the response of tobacco exports to dollariza-tion in both the short and the long run
vii. How have tobacco exports responded to dollarization in both the short and long run?
RER – real exchange rate. Source: own elaboration.
Justification and potential contribution of the study
Zimbabwe is a small open economy facing intensifying globalization and global effects. Hence, its macro price,
the real exchange rate, is becoming increasingly impor-tant. The real exchange rate signals how the domestic economy should adjust to the external environment, and maintain both domestic and external balance. Zimba-bwe is an agro-based economy, and therefore changes in the real exchange rate correspondingly determine the external competitiveness of agricultural export com-modities in the international market, as well as domestic production and domestic growth. Furthermore, changes in the real exchange rate affect input markets causing reduced output and reduced exports. Therefore, any agricultural export instability in Zimbabwe can lead to liquidity challenges, thereby affecting overall economic growth. The real exchange rate index is, however, an un-observable variable, and therefore a lot of care is needed to compute and interpret it.
LITERATURE REVIEW
The broad objective of this section of the study is to review literature on the response of exports to real ex-change rates, the volatility of the real exex-change rate, and their impacts on exports. This objective is achieved first by evaluating the evolution of Zimbabwe’s economic policies and economic performance from 1980 to 2019. Secondly, a review was carried out of literature concepts of real exchange rates, real exchange rate volatility, and real exchange rate equilibrium. The subsequent sections evaluate the different theories and empirical models ap-plied by researchers in estimating the response of agri-culture and overall trade to real exchange rates, real ex-change rate volatility and other factors. The final section discusses controversies in the debate on how exports respond to factors listed above.
The evolution of Zimbabwe’s macroeconomic policies and performance: 1980–2019
Zimbabwe attained independence in 1980. This led to lifting of international trade sanctions, thereby expos-ing the economy to international influence. Since 1980, the economy’s performance—as reflected by the GDP growth rate—was sporadically marked by high growth rates in the early 1980s. Then, in 1984, a recession struck due to drought. After the drought of 1984, stable eco-nomic growth was due to the removal of trade sanctions upon independence, and the opening up of external mar-kets which contributed to the increase in aggregate de-mand in the economy. In the 1990s, Zimbabwe launched
the economic structural adjustment program (ESAP) in-spired by the World Bank/International Monetary Fund. It was a reaction to the requirement for conditions of rapid economic growth (Mzumara, 2012). Flexible mar-ket-oriented macroeconomic policies were adopted. The monetary policy reform included market-based interest rates, parastatal commercialization and liberalization of the financial sector and of the labor market (Mzumara, 2012).
Pricing and marketing reforms were implemented in the agriculture sector. Constrains to private involvement in agriculture were removed together with the elimina-tion of a number of marketing board monopolies (Mum-bengegwi, 2002). The pre-arranged policy package had consequences on Zimbabwean economic growth. In 1992, the introduction of ESAP was accompanied by drought. The lowest GDP growth rate was recorded that year, as shown in Figure 1 above. The drought of 1992 affected agricultural production, and consequently had an impact on the overall economic growth. The poor performance of agricultural production severely af-fected the economy. The economic growth rate after the drought of 1992 was very low and highly erratic, slow-ing down to 1.9% per year on average between 1992 and 1996 (Sithole, 1996). The sharp decrease in GDP growth in 1997 and 1998 contributed to growing budget deficits caused by compensating war veterans with an amount equivalent to 3% of GDP (Buigut, 2015). After the introduction of the land reform in 2000, agricultural production greatly declined and so did economic growth.
Agriculture growth, however, recovered after the intro-duction of the dollarization policy in 2009. Economic growth also recovered, though it was later affected by tight liquidity conditions in the economy. However, the introduction of bond coins and notes in 2014 and 2016, respectively, failed to address the liquidity challenges facing the economy. With the introduction of local cur-rency and the discontinuation of the dollarization policy, the problem quickly resurfaced as inflation and decline in economic growth.
Concepts of RERs and RER volatility
The distinction between nominal and real can be ap-plied to both prices and exchange rates in the econo-my. Changes in nominal exchange rates do not neces-sarily mean changes in real exchange rates. A nominal exchange rate can depreciate while the real exchange rate appreciates if the country experiences high levels of inflation and other factors. According to Cheng et al. (2012), real appreciation of currency X is approximately equal to inflation of country X less nominal deprecia-tion of country X. All those who engage in internadeprecia-tional transactions are more interested in the real exchange rate than the nominal exchange rate. As a revisit, just like the GDP deflator and producer price index (PPI), RERs are another important macro-price. The GDP deflator measures the price level of all domestically produced goods and services in an economy. The real exchange rate measures how much domestically produced goods and services can be exchanged for goods and services
1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 -40 -20 0 20 40 60 80 100
Fig. 1. Nominal GDP growth rate (%) in Zimbabwe from 1980 to 2019
produced in another country (Ellis and Ellis, 2001). Real exchange rates are an important concept when analyzing international transactions and the impact of exchange rates on different policy variables (Loretan, 2005). Dar-vas (2012) mentioned that the real exchange rate serves many purposes, such as assessing the equilibrium value of currency, cost competitiveness, drivers of trade flows, and/or incentives for the reallocation of production be-tween tradable and non-tradable goods sectors.
The theory of RER and RER volatility
The real exchange rate is more important since it de-termines the country’s internal and external equilibrium position (Kipici and Kesriyeli, 1997). The real exchange rate helps assessing the equilibrium value of a currency, cost competitiveness, drivers of trade flows and/or in-centives for the reallocation of production between trad-able and the non-tradtrad-able goods sectors (Ellis, 2001). Appreciation of the real exchange rate discourages the exportable goods sector, mainly the natural resources sectors, while encouraging the importable goods sector, thereby discouraging economic growth (Nicita, 2013). Resources move to the non-tradable goods sector and strain the exportable goods sector (Mohamad and Ju-soff, 2008). Wong (2013) indicated that real exchange rates are volatile and never perfectly predictable. How-ever, in the long run, they are subject to a myriad of simultaneous pressures due to ups and downs in aggre-gate demand, inflation or deflation cycles, financial pan-ics, political crises, oil shocks, new technologies, abrupt changes in expectations, animal spirits, world pandem-ics and other factors.
The response of exports to RER, RER volatility and other factors
There are two major theories that explain the response of exports to factors that affect the export demand: the imperfect and perfect substitute models (Baier and Bergstrand, 2001). Trade equations were specified and estimated based on these two models. The fundamen-tal underlying assumption of the imperfect substitute model is that the analysis is set in a simple two-country model of foreign trade in which each country produces a single traded good which is an imperfect substitute for goods produced domestically in the other country (Os-man and Evans, 2005). Conversely, the perfect substitu-tion model assumes perfect substitutability between do-mestic and foreign goods. Under the perfect substitution
model, each country would be only an exporter or an importer of a traded good.
The model has attracted less attention in research as many researchers use the imperfect substitution model based on the Marshallian demand function (Camarero and Tamarit, 2003).
Payne and Mervar (2010) highlighted that the export demand function is specified according to the theory of export demand as a function of the importing country’s GDP, exported good’s own price and price of imperfect substitutes in the importing market (Cheng et al. 2012). They also pointed that in general, less significant fac-tors (such as unusual occurrences) can be incorporated into the model. The Bank of Uganda (2014) developed a two-country model of international trade based on the imperfect substitution model. The model assumes that the demand for a country’s exports is the function of real foreign income and relative prices. This relationship can be formulated as follows:
Xd = Xd (Yf, Pxf) (1)
where Xd is the demand for exports (Zimbabwean
to-bacco); Yf is foreign income (world income) proxied by
world GDP; Pxf = relative prices of exportable goods
abroad; and Pxf = P
E/EPf where PE is the price of
ex-portable goods in domestic currency per unit of foreign currency; E is the nominal exchange rate defined as the amount of local currency per unit foreign currency; and
Pf is the foreign price level in foreign currency. Empirical evidence of impact of RERs and RER volatility on exports
The evidence of impact of RERs and RER volatility on exports, trade position and economic growth remains controversial (Lubinga and Kiiza, 2013). Danquah (2014) examined the response of external trade to real exchange rate in Ghana. The study used the ARDL mod-el to analyze the export demand function. The variables were tested for stationarity using the Dickey–Fuller generalized least squares because non-stationary time series data can lead to spurious regression. A further test for cointegration was conducted to choose the appropri-ate model for the data. The results indicappropri-ated that in the long run, the response of exports to the real exchange rate was found to be negative. A 10% appreciation of the real exchange rate was found to reduce exports by 3.6% in the long run. The results suggested that an ap-preciation of the real exchange rate caused exports to
be uncompetitive in the international market. This re-sulted in deteriorating the trade balance in the long run. In the short run, the real exchange rate had an expected negative sign but was not significant. Hence, the study recommended the government to strengthen the coordi-nation between exchange rate and demand-side policies. In South Africa, Peter (2013) used the VECM model to assess the impact of real exchange rate volatility on the competitiveness of agricultural exports from 1980 to 2008. Real exports were a function of income from South Africa’s trading partners (mainly including Euro-pean countries), relative prices and real exchange rate volatility. The ADF test was used to test the stationarity of variables. The VECM was used and the authors won-dered whether the Engle–Granger test for co-integration and modeling co-integrated systems could have been ap-plied. However, they cited the shortcomings of the mod-el, namely that it could estimate up to one co-integrating relationship between the variables. The results revealed that real exchange rate volatility, foreign income, and relative prices had a positive impact on South African ag-ricultural exports in the short run, but in the long run, the real exchange rate had no significant impact. It is clear at this point that the impact of real exchange rates and real exchange rate volatility on exports remains controversial.
However, it is also essential to examine several is-sues that surround the theory and computation of nomi-nal or real exchange rates indexes. The absolute Pur-chasing Power Parity theory relies on the assumption of the law of one price. The law of one price always im-plies integrated competition markets. However, the spot price of a given commodity will not necessarily be equal in different locations at a given time because of the in-ability to shift commodities instantaneously from one location to another. The basket of commodities tends to differ between countries, and the price measures across countries are unlikely to be constructed in terms of ab-solute amounts.
Weighted average bilateral nominal exchange rates are used to assess how the exchange rate of the domestic currency appreciated or depreciated relative to the ex-change rate of trading partners compared to a base year. Some researchers, however, refer to the computed index as an effective nominal exchange rate or multilateral nominal exchange rate which might be confusing. This methodological ambiguity has often led researchers to use these terms interchangeably which is sometimes puzzling and difficult to understand.
RESEARCH METHODOLOGY
The objective of this section is to present research meth-ods and techniques used in evaluating the response of tobacco exports in Zimbabwe to RERs, RERV and other factors. This broad objective is achieved first by presenting the conceptual framework of the response of exports to these variables. The second part describes the steps used in the computation of the RER index. The paper then presents the process of generating the RERV index using GARCH and ARCH models. Thereafter, it provides a model for estimating the response of tobacco exports to RERs and RERV indexes and other factors.
The RER is defined as the nominal exchange rate adjusted for price differential between trading partners. Mathematically, it can be presented as:
Pf
RER = e Pd (2) where RER is the real exchange rate index, e is the bilat-eral nominal exchange rate, Pf is the foreign price level
and Pd is the domestic price level. The decrease in the
index based on the above formula represents an appreci-ation of the RER. The home currency appreciates in real terms. An increase in the index represents a deprecia-tion of the RER. The home currency depreciates in real terms (Kipici and Kesriyeli 1997; Opoku-afari 2004). RER can drop due to three events, i.e. a decline in Pf
, an increase Pd, and/or a decline in e. All these events
can lead to a depreciation of the real exchange rate of the home currency. The RER can increase if Pf
increas-es, Pd falls and e increases, causing the appreciation of
the home RER. However, simultaneous and differential changes in these three factors can lead to indeterminate changes to RERs.
Methods of computing RER, RER volatility and the exports response function
A chain-linked multilateral real exchange rate index ag-gregated from individual bilateral real exchange rates with the weighted values for the corresponding years is constructed as follows:
MRERt = (BRER1t) W1t + (BRER2t) W2t … +
(BRERNt) WNt
where: BNER is defined as the bilateral nominal ex-change rate for Zimbabwe with country i at time t. BRERt is BNER adjusted for price differentials using
GDP deflators. Wit is the weight of country i at time t.
The autoregressive process generates the real exchange rate volatility based on the chain-linked multilateral real exchange rate index as follows:
∆RERt = α0+ ∑pi∂i∆RERt−1 + ut (3)
where α0 is a constant, ∂i are coefficients of lagged values
of RER and ut|Ωt − 1~N(0, δ2) is the error term which is
normally distributed with a zero mean and variance δ2.
The set of available information at time t–1 is denoted by |Ωt − 1|. Tabar-Hashemi (2009) notes that variance
is allowed to vary over time, and the idea behind the ARCH model is to characterize the changes in variance over time. Variance is assumed to be captured by the following autoregressive process:
δ2
t= ƛ0 + ∑pi=1Øi μ2t–i (4)
where δ2
t is the conditional variance of the real exchange
rate, μ2
t–i are the squared residuals derived from equation
(4), and Øi are the parameters to be estimated. Øi ≥ 0,
this condition is necessary to ensure that the predicted variance is always positive. In equation (4), the current levels of misalignment are determined by the misalign-ment of the previous levels. Therefore, low or high peri-ods of movements tend to continue. On the other hand, to extend the ARCH model, a GARCH (p, q) process will then be introduced in which δ2
t becomes a function
of both μ2
t−i and lagged values of itself. The ARCH term,
or the number of autoregressive lags denoted by p and q, is the GARCH term (or specifies the number of moving averages used). The conditional variance is then esti-mated as shown below:
δ2
t = λ0 + Σpi=1Øiμ2t–i = Σqi=1δi σ2t–i (5)
To have a positive variance, all coefficients in equa-tion (5) must be positive.
The derived export response function based on the imperfect substitution model is applied to the tobacco export response model for Zimbabwe. The model is log-arithmized, except for real exchange rate volatility and dollarization because real exchange volatility can take negative values whereas dollarization is a dummy varia-ble. Thus, the tobacco export response model reduces to:
ln (EXPTt) = b0 + b1lnRWGDPf t + b2 lnRER t +
b3 lnTOTt + b4RERVt + b5 DOLt +et
Where EXPT is the tobacco export volume during the period t, RWGDPf t is the real world gross domestic
product or import intensity, RERt is the real exchange rate
within the sample period, TOTt is a measure of the
rela-tive price level of export/import prices (proxied for terms of trade), RERVt is a measure of the real exchange rate
volatility, DOLt is dollarization and et is the error term.
Economic agents take time to respond to new eco-nomic conditions. As a consequence, state interven-tions have different impacts in the short run because of policy time lags. In the long run, economic agents have enough information and time to adjust to market information and conditions. As a result, export activi-ties carried out by rational economic agents will respond to policy changes in the long run, yet economic agents might not respond in the short run. The error correction term is expected to be negative in the short and long run. However, the negative sign on the error correction term (ECT) indicates that the model adjusts back to the equilibrium after an exogenous shock.
Estimation of RER and RER volatility indexes
This section seeks to compute the real exchange rate and real exchange rate volatility indexes. To achieve this broad objective, the chapter starts by presenting time series data and its sources. The second section computes the real exchange rate index using a time-variant weighting system and GDP deflators for each trading partner. Thirdly, the real exchange rate volatility index is generated using the GARCH and ARCH mod-els. The section concludes by discussing the computed indexes.
Table 2. Prior expectation of the response of tobacco exports
to independent variables
Independent variables
Expected
signs Expected signs Short run Long run
Real world GDP (LNRWGDP) – +
Real exchange rate (LNRER) – +
Terms of trade (LNTOT) – +
Real exchange rate misalignment
(RERM) + –
Dollarization – +
-Data and data sources
This study uses published secondary annual time series data for Zimbabwe, spanning from 1980 to 2019. The data was collected during a period where Zimbabwe has had major inflationary episodes and changes of curren-cy. This compromises the quality, consistency and reli-ability of data used in this study. The data for weights, exchange rates, real world GDP, and GDP deflators used in computing the real exchange rate index was collected from the World Bank (WB), the International Monetary Fund and the United Nations commodity trade. Data on tobacco exports was collected from the Tobacco Mar-keting Board. The major problem with this data is the change of currency in 2009 when the country dollarized its economy. Up until 2008, the exchange rate was spec-ified for the Zimbabwean dollar (per unit of foreign cur-rency). After March 2009, it was expressed as USD per unit of foreign currency. The currency has now changed back to Zimbabwean dollar. These changes are confus-ing and complicated, makconfus-ing both a static and dynamic analysis and interpretation very difficult. Most develop-ing countries face data collection, compilation, and pro-cessing challenges, particularly during inflationary peri-ods when prices are dramatically changing (Mohamad and Jusoff, 2008).
Computation of the RER index
The process of computing the real exchange rate in-volves several interrelated steps. The first one is the se-lection of bilateral nominal exchange rates. This study selected ten bilateral nominal exchange rates for major trading partners comprising more than 50% of Zimba-bwean total foreign trade. The three-year time-depend-ent weights for ten selected major trading partners were calculated, as reported in Table 3 below.
DISCUSSION OF RESULTS, POLICY IMPLICATIONS AND CONCLUSIONS
This section explores estimation of results, policy impli-cations and conclusions. In order to achieve this broad objective, the first part discusses the real exchange rate and real exchange rate volatility indexes. The second section discuses short-run and long-run tobacco ex-port demand elasticities and dynamics. The final sec-tion presents policy implicasec-tions and recommendasec-tions. The chapter concludes with a summary, conclusion and identification of areas of further research.
Table 3. The evolution of weights for Zimbabwe’s major trading partners in percentages (%)
Country Year 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015 2018 Germany 11.93 7.4 8.99 8.03 9.9 3.4 6.8 2.53 3.09 2.8 2.3 Japan 7.82 9.47 9.69 9.07 7.42 2.71 4.65 2.15 1.17 2.4 0.9 China 4.48 2.74 1.68 4.96 0.59 0.14 8.89 10.7 19.88 19.9 18.03 France 2.85 2.53 3.07 3.15 2.45 0.72 0.76 0.86 1.03 1.6 1.98 South Africa 35.61 37.35 49.13 39.47 41.47 20.61 56.55 64.69 54.81 56.3 58.16 UK 11.21 16.35 11.61 12.46 13.5 3.39 4.16 4.21 2.45 2 2.45 Zambia 4.55 3.3 2.85 5.04 5.86 3.79 7.49 5.09 10.26 8.6 9.6 U.S. 7.7 11.88 7.31 8.54 6.6 1.75 4.99 3.93 2.09 2.12 2.08 Netherlands 7.84 4.2 4.26 4.97 6.87 1.94 2.42 4.11 2.11 2.12 2.1 Italy 5.98 4.78 1.39 4.26 5.35 1.21 3.38 1.72 3.09 2.16 2.4 Total 100 100 100 100 100 100 100 100 100 100 100
Discussion of estimated RERs and RER volatility indexes
The RER index has been depreciating since 1980, though at a low rate. The RER index depreciated by 39%. This implies that the Zimbabwean dollar depreciated in real terms during that period. The introduction of ESAP in the 1990s marked the beginning of RER index fluctua-tions. The real exchange rate responded to changes in the exchange rate policy in 1992 and depreciated by 77% during that period. However, the depreciation was par-ticularly severe in 2000–2008, during the land reform
program. The decline in agricultural production affected the supply of foreign currency and caused instability in the foreign exchange market. The RER index greatly de-preciated by 545%. The value of the Zimbabwean dollar declined sharply against the value of trading partners’ currencies.
It can be shown that the volatility of the real exchange rate was very low in the period just after independence. This can be attributed to the pre-independence fixed ex-change rate system adopted by the Zimbabwean govern-ment in 1980. The small fluctuations in the exchange
1980 1982 1984 1986 1988 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 0 100 200 300 400 500 600 700 800 RER INDEX
Fig. 2. Zimbabwean real exchange rate index in 1980–2019
19750 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2 4 6 8 10 12 14 16 RERV INDEX year
rate imply that it was stable and that the risk associated with its fluctuations was very low. Also, the firms faced only a small risk when entering the international market. Fluctuations in the real exchange rate started in 1992 when ESAP was introduced. This is evidenced by Fig-ure 2 above, and can be attributed to the fact that the government liberalized the fixed exchange rate system, shifting to a market-determined exchange rate. In that period, low volatility of the exchange rate could have promoted foreign trade.
The period from 2000 to 2008 was marked by a highly volatile real exchange rate, as evidenced by the corresponding index. This is the period when Zimba-bwe launched a land reform program known for hav-ing destabilized agricultural production, a pillar of the Zimbabwean economy (Scoones et al., 2011). The high real exchange rate volatility over that period can be con-nected to hyperinflation, large balance of payments defi-cits and great depreciations of the Zimbabwean dollar. The high exchange rate volatility over that period could have affected Zimbabwe’s foreign trade. This is because domestic firms would not be willing to participate in foreign markets due to a high risk which can have an adverse effect on their profit margins. Agricultural and non-agricultural export commodities could have been affected by the random movements of the exchange rate. Under the dollarization policy, volatility of the real ex-change rate was low compared to the pre-dollarization period. This is because dollarization of the Zimbabwean economy tamed large depreciations of the domestic cur-rency and the rampant hyperinflation. The adoption of the United States dollar as the anchor currency in the economy reduced the volatility of the exchange rate to
low levels. A low volatility of the exchange rate encour-ages domestic firms to participate in foreign markets due to reduced risks associated with exchange rate fluc-tuations. The recent introduction of bond notes as the local currency has resulted in the rise of exchange rate volatility on the local market.
Results and discussion of the tobacco exports demand function
The results of the correlation test for tobacco exports, real world GDP (WGDP), real exchange rate (RER), terms of trade (TOT), real exchange rate volatility (RERV) and dollarization are summarized in Table 4 below.
The results of the correlation test reveal that the vari-ables in the model are not correlated.
The variables were further tested for stationarity us-ing the unit root tests. The Augmented Dickey–Fuller (ADF) test is used to check if the variables are stationary at levels 1(0). If the probability is above the 5 % level of significance, it indicates that the variables are not stationary. Conversely, if the variables are below the 5% level of significance, they are stationary. Therefore, the null hypothesis tested is that H0: δ = 0 (i.e. that the
variable is non-stationary), whereas the alternative hy-pothesis is that H1: δ ≠ 0 (i.e. the variable is stationary)
(Odili, 2015). All the variables were found to be non-stationary at levels. Stationarity is indicated by their re-spective probabilities which are all above the 5% level of significance, as shown in Table 5 below.
As shown by the results, the null hypothesis that the variables under consideration have a unit root could not be rejected. This means that the variables are not sta-tionary at levels, and are not integrated of order zero
Table 4. Correlation tests
LNEXPORTS LNRWGDP LNRER LNTOT RERV DN
LNEXPORTS 1 0.176 0.381 –0.103 0.0920 0.0351 LNRWGDP 0.176 1 0.136 –0.072 0.056 –0.435 LNRER 0.381 0.136 1 –0.130 –0.096 –0.1107 LNTOT –0.103 –0.072 –0.130 1 –0.088 0.0897 RERV 0.092 0.056 –0.096 –0.088 1 –0.0340 DN 0.035 –0.435 –0.110 0.089 –0.034 1
I(0). Table 5 above indicates that even with intercept and trend, the variables under study are still not station-ary at levels and are not I(0). Since the variables are not stationary at levels, all variables are differenced at first level, and further stationary tests are carried out. Sta-tionarity tests for unit roots on differenced data revealed the non-existence of unit roots. The differenced series were found to be stationary. Table 6 below presents the respective probabilities which are all below 5%.
The null hypothesis that the variables have a unit root is rejected in favor of the alternative hypothesis which states that the variables do not have a unit root. Therefore, all variables in the model are stationary at first difference.
Since the level of data integration is known, a coin-tegration test can be carried out to check for a long-run
relationship. Cointegration tests seek to establish any long-run relationship between variables. Given that this data is now stationary at first difference, the Johansen maximum likelihood criteria is used to find the number of cointegrating equations in the model. The Johansen method is used because the variables are integrated of the same order. The cointegration test based on the Jo-hansen maximum likelihood procedure was carried out to establish a long-run relationship, and revealed the existence of such a relationship between the variables. The trace statistic tests the null hypothesis of r cointe-grating relations against the alternative of n cointegrat-ing relations (n is the number of variables in the model where r = 0, 1, 2, n–1). The maximum eigenvalue tests
the null hypothesis of r cointegrating relations against the alternative of r+1 cointegrating relations for r = 0, 1, 2, n–1 (Odili, 2015). If there is at least one
cointegrat-ing equation in the model, both in trace statistic test and in the maximum eigenvalue test, the null hypothesis is rejected at 5% in favor of the alternative hypothesis. Ta-ble 7 below summaries these results.
In both (trace and maximum eigenvalue) tests for cointegration, the null hypothesis that no cointegra-tion exists among variables is rejected at the 5% level of significance. The alternative hypothesis (that at least 1 cointegrating equation exists among variables) is ac-cepted. This indicates that there is a long-run relation-ship or association between the variables. Both tests indicated that there is 1 cointegrating equation at the 5% level of significance.
Specification and estimation of the VECM
Since the data revealed the absence of autocorrelations among variables and demonstrated that all variables are integrated of order 1 and have at least one cointegrating relationship, it is now possible to specify and estimate a vector error correction model and to compute both long-run and short-run elasticities and the error correc-tion term. A VECM is recommended since variables are integrated of the same order and are cointegrated (Obayelu and Salau 2010). According to Odili (2015), the VECM is used if a set of variables is found to have one or more cointegrating vectors, since it incorporates both short-run and long-run effects of variables.
In order to specify the VECM, lag lengths of both the independent (p) and independent variables (q) must be known. The optimal lag length of variables includ-ed in the VECM model is obtaininclud-ed from the selection
Table 5. ADF-based unit root tests at levels I (0).
LNEX-PORTS, LNRWDGP, LNRER, LNTOT and RERV
Variables With intercept
With intercept and trend I(0) Prob. I(0) Prob.
LNEXPORTS 0.4502 0.7681
LNRWGDP 0.8098 0.8652
LNRER 0.0943 0.0554
LNTOT 0.2198 0.1997
RERV 0.1036 0.2465
Source: Eviews Software Computation 2020.
Table 6. ADF-based unit root tests for differenced data, I(1).
LNEXPORTS, LNRWDGP, LNRER, LNTOT and RERV
Variables With intercept
With intercept and trend
I(I) Prob I(I)Prob
LNEXPORTS 0.0000 0.0000
LNRWGDP 0.0002 0.0003
LNERER 0.0000 0.0000
LNTOT 0.0001 0.0002
RERV 0.0000 0.0000
criteria based on the Akaike’s Information Criteria (AIC), Schwartz-Bayesian Criteria (SBC) and the Han-nan–Quinn information criterion. These models indicate the lowest value; and the lower the value, the better the model. If more than one criterion suggests the same number of lags, that lag number is selected. Table 8 be-low summarizes the result of the estimate.
The table indicates that the optimum lag length is one for the specified vector error correction model. We can now specify the VECM for the estimation of the short run, long-run and the error correction term of the tobacco export demand function.
Given the lag information, the VECM is also test-ed for its statistical properties of serial correlation,
Table 7. Cointegration test for LNEXPORTS, LNR WDGP, LNRER, LNTOT, RERV and DN
Hypothesized Trace 0.05
Unrestricted cointegration rank test (trace)
No. of CE(s) Eigenvalue Statistic Critical value Prob.**
None * 0.918058 149.0147 95.75366 0.0000 At most 1 0.532483 66.45733 69.81889 0.0900 At most 2 0.471978 41.36677 47.85613 0.1773 At most 3 0.268750 20.29240 29.79707 0.4032 At most 4 0.244401 9.963423 15.49471 0.2835 At most 5 0.021444 0.715337 3.841466 0.3977 Hypothesized Max-Eigen 0.05
Unrestricted cointegration rank test (maximum eigenvalue)
No. of CE(s) Eigenvalue Statistic Critical value Prob.**
None * 0.918058 82.55736 40.07757 0.0000 At most 1 0.532483 25.09056 33.87687 0.3789 At most 2 0.471978 21.07437 27.58434 0.2717 At most 3 0.268750 10.32898 21.13162 0.7133 At most 4 0.244401 9.248086 14.26460 0.2661 At most 5 0.021444 0.715337 3.841466 0.3977
The trace test indicates 1 cointegrating eqn(s) at the 0.05 level. * Denotes rejection of the hypothesis at the 0.05 level. Source: Eviews Software Computation 2020.
Table 8. Lag length selection criteria. LNEXPORTS, LNRWGDP, LNRER, LNTOT, RERV and DN
lag LL LR df p FPE AIC HQIC SBIC
0 –1 773.89 6.2e+40 108.115 108.267 108.568
1 –1 653.41 240.96 25 0.000 7.0e+38 102.328* 102.005* 103.811* 2 –1 607.98 90.862* 25 0.000 6.5e+37* 103.09 103.345 103.835 Source: Eviews Software Computation 2020.
heteroscedasticity, normality and stability. The model is tested for serial correlation using the Breusch–Godfrey serial correlation LM test. The results are reported in Table 9 below.
The statistics reveal the absence of autocorrelation by rejecting the existence of autocorrelation based on the Breusch–Godfrey serial correlation LM test. The F-statistic and chi-squared probabilities are all above 5%, indicating that the null hypothesis on the absence of se-rial correlation cannot be rejected. Based on the above results, the model has no serial correlation. The Q-statis-tic also confirms the absence of serial correlation in the estimated model.
Furthermore, the model was checked for the presence of heteroscedasticity using the Breusch–Pagan–Godfrey test. If the probability of the F-statistic is above 5%, the null hypothesis on the absence of heteroscedasticity is accepted. Conversely, if the probability is below 5%, the alternative hypothesis on the presence of heteroscedas-ticity in the model is accepted. Table 10 below summa-rizes the results.
The statistics reveal the absence of heteroscedastic-ity based on the Breusch–Pagan–Godfrey test. The re-sults indicate that the F-statistic’s probability is above the 5% level of significance, indicating that the null
hypothesis on the absence of heteroscedasticity in the model cannot be rejected. Hence, the model is free from heteroscedasticity.
The vector error correction model is further tested for normality to find if the model is normally distrib-uted. If the probability is below 5%, the null hypoth-esis on the model’s normal distribution is rejected. Conversely, if the probability is above 5%, the null hy-pothesis is accepted. The distribution of the vector er-ror correction model is presented in the appendix. The probability (0.62459) is above the 5% level of signifi-cance. Thus, the null hypothesis on the model’s normal distribution cannot be rejected; the model is normally distributed at the 5% level of significance. Finally, the model was tested for stability. These tests are used to examine the stability of long-run parameters together with the short-run movements for the equations (Alimi, 2014). The stability of the estimated error correction model should be empirically investigated (Pahlavani et al., 2005). The stability of the model is tested us-ing the cumulative sum (CUSUM) and cumulative sum squares (CUSUMSQ) tests. If the plots of CUSUM and CUSUMSQ statistics are within the critical bounds at the 5% level of significance, the null hypothesis that the coefficients in the model are stable is accepted. However, if the plots of CUSUM and CUSUMSQ sta-tistics are outside the critical bounds at the 5% level, the null hypothesis that all coefficients in the model are stable is rejected (Berthou, 2008). CUSUM and CUSUMSQ statistics are presented in the appendix. The plot of cumulative sum (CUSUM) and cumulative sum squares (CUSUMSQ) stays within the critical 5% boundary. Hence, the null hypothesis that the regres-sion equation is correctly specified cannot be rejected. Therefore, this further confirms the existence of a long-run relationship between variables and the stability of the model’s coefficients. The above results therefore imply that the model can be used for hypothesis test-ing and forecasttest-ing, since the vector error correction model is normally distributed with classical statistic properties.
The VECM specified below summarizes the estimat-ed long-run elasticities. The estimatestimat-ed long-run cointe-grating equation is as follows:
ln EXPORTS = –0.056819 + 77.6279 ln RWGDPt-1 –
2.140 ln RERt-1 – 72.10862 ln TOTt-1 + 2.40 RERVt-1 +
0.878164 Dnt-1
Table 9. Serial correlation test
Breusch–Godfrey serial correlation LM test F-statistic 2.843373 Prob. F(1,24) 0.1047 Obs*R-squared 3.495511 Prob. Chi-Squared(1) 0.0615 Source: Eviews Software Computation 2020.
Table 10. Heteroscedasticity test
Heteroscedasticity Test: Breusch–Pagan–Godfrey F-statistic 1.222551Prob. F(11,21) 0.3319 Obs*R-squared 12.88276 Prob. Chi-Squared(11) 0.3011 Scaled explained SS 9.024176Prob. Chi-Squared(11) 0.6197 Source: Eviews Software Computation 2020.
The estimated long-run cointegrating equation re-vealed that the coefficients of RER, terms of trade and real exchange rate volatility were significant in the long run while those of dollarization and real world GDP were not. The coefficients of variables used in the coin-tegration equation are presented in the appendix.
The VECM further estimates short-run elasticities, as shown in Table 11 below together with the dynamics of the vector error correction model.
The model is properly fitted with 87.6% adjusted R-squared; the Prob(F-statistic) is also significant at the 1% level. The results indicate that the model is specified correctly.
World GDP
The results of the tobacco export demand function re-vealed that tobacco exports were found to have a nega-tive insignificant response at a level of 32% to a 10% increase in real world GDP in the short run, and a posi-tive insignificant response at a level of 77% to a 10%
increase in real world GDP in the long run. The null hypothesis that the response of tobacco exports to an in-crease in import intensity is negative in the short run and positive in the long run is rejected since real world GDP was found to be insignificant in both the short and long run. The results suggest that the response of tobacco ex-ports to an increase in global economic downturn or re-cession is negative and insignificant in the short run, and hence fails to explain any variation in tobacco exports. Again, the response of tobacco exports to an increase in real world GDP is positive but insignificant to explain the variation in tobacco exports in the long run.
Real Exchange Rate
Tobacco exports are found to have a positive response at a level of 49% to a 10% depreciation in the RER in the short run. The alternative hypothesis that tobacco exports respond positively to RER depreciation in the short run is accepted. Tobacco exports have a negative response at a level of 24 % to a 10% appreciation or
Table 11. Results of short-run elasticities and dynamics of the VECM
Header Coefficient Std. error t-Statistic Prob. LNEXPORTS (–1) –0.442579 0.069106 –6.404343 0.0000*
LNRWGDP (–1) –3.174855 –27.86504 –0.113937 0.9102
LNRER (–1) 4.92E–12 2.24E–12 2.192393 0.0379**
LNTOT (–1) 9.129417 3.000867 3.042260 0.0055*
RERV (–1) –3.26E–11 8.74E–12 –3.734956 0.0010*
DON (–1) –4.605568 1.653934 –2.784614 0.0101**
Error correction term (ECT) –0.194986 0.0000
–0.194986 0.03745 –5.20679 0.0000*
R–squared 0.903540 Mean dependent var 0.207273
Adjusted R–squared 0.876531 S.D. dependent var 3.207633 S.E. of regression 1.127105 Akaike info criterion 3.284399 Sum of squared resid 31.75914 Schwarz criterion 3.647188 Log likelihood –46.19258 Hannan–Quinn criter. 3.406466
F–statistic 33.45335 Durbin–Watson stat 1.380898
Prob(F–statistic) 0.000000 * At the 1% level of significance. ** At the 5% level of significance.
decrease in the RER in the long run. The null hypothesis that tobacco exports respond negatively to an apprecia-tion of the RER index is accepted. This implies that the depreciation of the RER has resulted in the deprecia-tion of the Zimbabwean dollar and then of the United States dollar in real terms against currencies of trading partners. The depreciation of the local currency makes domestically produced goods cheap compared to those produced abroad. This improves export competitiveness and could be the reason behind the positive impact on tobacco exports in the short run. Therefore, a deprecia-tion of the RER in the short run has a positive impact on the demand for tobacco exports in the international market.
Huchet-Bourdon and Korinek (2011) mentioned that short-run effects of exchange rate levels on trade do not normally follow a specific pattern. However, short-run effects can be tested for the existence of a J– Curve effect which means that the an increase or de-preciation in the real exchange rate is associated with a deterioration in the trade balance in the short run. This is because immediately following RER deprecia-tion, the volume of trade will remain unchanged due to the recognition of the adjustment period by importing countries (Ali et al., 2015). However, the current study of the response to RER does not confirm the presence of the J–Curve phenomenon in the short run. Neverthe-less, in the long run, a negative response of tobacco exports to RER appreciation or decrease leads to an appreciation of the Zimbabwean dollar and then of the United States dollar in real terms. An appreciation of the domestic currency increases export prices, lead-ing to a loss of external competitiveness of exported goods. This could be the reason why the appreciation of the RER had a negative impact of 24% on tobacco exports in the long run. The same long-run effect was found by Danquah (2014): the RER was negatively related to exports in Ghana. However, in the current study, the long-run response of tobacco exports to RER confirms the presence of the J–Curve effect. It states that the long-run appreciation of the domestic currency in real terms harms exports due to the recognition of the adjustment period by importing countries search-ing for alternatives where goods are cheap (Yazici, 2006). Hence, the appreciation of the Zimbabwean dollar and then of the USD in real terms could have negatively affected the demand for tobacco exports in the international market.
Terms of Trade (TOT)
Tobacco exports are found to have a positive response at a level of 91% to a 10% increase in terms of trade (TOT) in the short run. The alternative hypothesis that tobacco exports respond positively to an increase in TOT in the short run is accepted. In the long run, the response of tobacco is negative at a level of ca. 721% to a 10% re-duction in TOT. The alternative hypothesis that tobacco exports respond negatively to TOT reduction in the long run is accepted. A positive commodity-market-specific shock or a positive world demand shock can increase the demand for commodities in global markets and bid up their prices around the world. Favorable export pric-es encourage crop production by firms and individual farmers, thereby increasing tobacco export volumes to the international market. Again, it implies that in the short run, tobacco farmers are highly responsive to an increase in world commodity prices. Therefore, an in-crease in export prices could act as an incentive for to-bacco production in the short run. In the long run, the negative response of tobacco exports to a reduction in TOT can be the result of a negative commodity-market-specific shock or a negative world demand shock that can reduce the demand for commodities in the world markets. As a result, commodity prices in the world markets decline. Unfavorable export prices discourage firms and individual farmers from crop production in developing countries like Zimbabwe, thereby reduc-ing the volume of tobacco exports to the international market.
Real exchange rate volatility
Tobacco exports are found to have a negative response at a level of 33% to a 10% increase in RER volatility in the short run. The alternative hypothesis that tobacco exports respond negatively to an increase in RER vola-tility in the short run is accepted. In the long run, to-bacco exports respond positively at a level of ca. 22% to a 10% increase in RER volatility. The alternative hypothesis that tobacco exports respond positively to an increase in RER volatility in the long run is accept-ed. The results suggest that volatility of real exchange rates (especially including overvaluations attributable to unsustainable monetary and fiscal policies) have had negative impacts on tobacco export performance. The increase in RER volatility could have made tobacco ex-ports uncompetitive in the international market due to an overvalued local currency. Further deviation of the
real exchange rate from its equilibrium values also in-dicates that the exchange rate was very volatile, thereby discouraging tobacco exports in the short run. A higher degree of exchange rate volatility could reduce trade volumes because higher fluctuations in the exchange rate pose a high risk on profits. Risk-averse firms and risk-neutral producers reallocate resources away from high-risk global markets to low-risk domestic markets (Oiro and Otieno, 2015). This therefore means that the greater the volatility of the exchange rate in the short run, the smaller the exports of more risk-averse ex-porters in the short run. Indeed, some risk-averse firms could have exited the foreign market due to exchange rate fluctuations. Further, it implies that the importing countries in the world market are highly responsive to short-run movements in the real exchange rate. Hence, RER volatility can affect demand for tobacco exports in global markets.
Dollarization
Tobacco exports are found to have a negative response at a level of 46% to the lack of dollarization in the short run, and a positive insignificant response at a level of 88% to the introduction of dollarization in the long run. The null hypothesis that the response of tobacco exports to the introduction of dollarization is negative in the short run and positive in the long run was accepted in the short run and rejected in the long run. This means that the introduction of dollarization does not have a signifi-cant short-run impact on tobacco exports. This might be because in the short run, economic agents were adjust-ing to the new policy intervention. Hence, the variation in exports could be explained by other factors. How-ever, in the long run, the introduction of dollarization is not significant and hence cannot explain the variations in tobacco exports.
Again, it implies that the variation in tobacco ex-ports can be attributed to other factors in the economy. However, Thaver and Bova (2014) found the same result in the short run but a different one in the long run. Dollarization had a negative impact on Ecuador’s exports in both the short and long run. The decrease in the lagged values of tobacco exports was found to have a negative response to tobacco exports. A 10% decrease in the lagged values of tobacco exports was found to have a 44% negative effect on tobacco ex-ports. In the very short run, tobacco export volumes can be inelastic to changes in tobacco export prices.
However, a short-run period could be enough for to-bacco exports to adjust to changes in toto-bacco ex-port prices. A decline in tobacco exex-port prices due to a decrease in world demand for tobacco could reduce the output and export of tobacco. It also implies that when other firms exit the foreign sector due to lower tobacco export prices, other firms follow suit, lead-ing to reduced output for the export market. The co-efficient on the error correction term (ECT) is found to be significant and negative and not lower than –2, which further confirmed the existence of a long-run relationship between the variables (Odili, 2015). This indicates that the model adjusts quickly back to the equilibrium at a rate of 195% following an exogenous shock within a year.
Policy implications of the estimated results
The real exchange rate appreciation was found to nega-tively affect tobacco exports in the long run. To restore competitiveness of agriculture-based commodities like tobacco under the dollarization policy, the government can devalue the real exchange rate. Devaluation of the real exchange rate can restore competitiveness of trad-able goods like tobacco through adjusting for inflation. This is achieved by lowering production or wage costs and improving productivity without lowering the nomi-nal exchange rate value. The above can reduce prices in input markets, lowering the production cost for both individual farmers and firms engaged in the production of tobacco for export markets. The reduction in terms of trade was found to have a negative effect on tobacco exports in the long run. The government can improve the terms of trade by increasing tax on importable goods and simultaneously putting in place policies that en-courage local industry growth. The real exchange rate volatility was found to depress tobacco exports in the short run. Hence, the government can adopt other mech-anisms to reduce uncertain movements of the exchange rate. For example, the government can establish an op-erating environment which enables hedging of export sales through forward sales in foreign currency. Also, the dollarization policy was found to have a negative impact on tobacco exports in the short run. Therefore, the government can increase domestic interest rates to attract foreign investment. The resulting capital inflows can ease liquidity challenges in the economy and al-low the farmers to access funds to finance agricultural operations.
Fig. 4. Normal distribution of the vector error correction model
Source: Eviews Software Computation 2020.
Fig. 5. Cumulative sum (CUSUM) test
Source: Eviews Software Computation 2020.
Fig. 6. Cumulative sum squares (CUSUMSQ) test
Table 12. Results of long-run elasticities of the vector error
correction model
Variables: Cointegrating Equation
LNEXPORTS (–1) 1.000000 LNRWGDP (–1) –77.6279 (74.45627) [–1.04] LNRER(–1) 2.14E–10** (2.3E–11) [9.36960] LNTOT(–1) 72.10862** (8.29232) [8.69583] RERV(–1) –2.40E–10** (2.2E–11) [–11.0776] DON(–1) –0.878164 (0.94282) [–0.93142] C 0.056819
** Significant at the 5% level.
Source: Eviews Software Computation 2020.
CONCLUSION AND AREAS FOR FURTHER RESEARCH
Under the dollarization regime, devaluation of the real exchange rate can restore competitiveness of tradable goods like tobacco. This can be achieved by lowering production or wage costs and improving productiv-ity without lowering the nominal exchange rate val-ue. The introduction of the local currency resulted in real exchange rate depreciation and boosted inflation. Therefore, the government is advised to first address the macroeconomic fundamentals before introduc-ing local currency. Otherwise, the economy will self-dollarize, rendering de-dollarization efforts impracti-cal. The reintroduction of the multicurrency regime would be an alternative route but should be supported by high export growth to ease liquidity challenges in the economy. It is, however, important to craft policies that stabilize the exchange rate, encourage investment
and industry production for any monetary policy to be effective. There is need to carry out further research on exchange rate misalignment and investigate its ef-fect on agricultural export commodities in Zimbabwe. Further, there is also need for research on the impact of real exchange rate and real exchange rate volatility on economic growth in Zimbabwe. This will help the government to design exchange rate policies that can improve external sector performance and drive overall economic growth.
REFERENCES
Ali, A., Johari. F., Alias, M. (2015). On The J-Curve Theo-ry: The Literature and Further Proposed Improvements. Economy, 2(4), 64–70.
Danquah, B.A. (2014). The Effects of Exchange Rate on Ghana’s External Trade. British J. Econ. Fin. Manag. Sci., 9(1), 48–68.
Alimi, R.S. (2014). ARDL Bounds Testing Approach to Coin-tegration: A Re- Examination Of Augmented Fisher Hy-pothesis In An Open Economy. Asian J. Econ. Model., 2(2), 103–114.
Baier, S.L., Bergstrand, J.H. (2001). The Growth of World Trade- Tarrifs, Transport Costs, and Income Similarity. J. Int. Econ., 5(53), 1–27.
Bank of Uganda (2014). Does Real Effective Exchange Rate Volatility matter for Uganda’s Flowers Exports? A Cointe-grated VECM Approach, Kampala.
Buigut, S. (2015). The effect of Zimbabwe’s Multi-Currency Arrangement on Bilateral Trade: Myth Versus Reality. Int. J. Econ. Fin. Iss., 5(3), 690–700.
Camarero, M., Tamarit, C. (2003). Estimating exports and im-ports demand for manufactured goods: The role of FDI. Jaume I University.
Cheng, K.M., Kim, H., Thompson, H. (2012). The real ex-change rate and the balance of trade in US tourism. Int. Rev. Econ. Fin., 25, 122–128.
Darvas, Z. (2012). A tale of three countries: recovery after banking crises (No. MT-DP-2012/2). IEHAS Discussion Papers.
Ellis, L., Ellis, L. (2001). Measuring the real exchange rate: Pitfalls and practicalities (pp. 2001-2004). Reserve Bank of Australia.
Government of Zimbabwe (2016). The 2016 National Budget
Statement, Harare. Retrieved Oct 27th 2016 from: www.
dpcorp.corp.co.zw/assets
Kipici, A., Kesriyeli, M. (1997). The Real Exchange Rate Definitions and Calculations. Central Bank of the Repub-lic of Turkey.
Loretan, M. (2005). Indexes of the Foreign Exchange Value of the Dollar, Washington DC.
Lubinga, M.H., Kiiza, B. (2013). Exchange Rate Uncertainty and Bilateral Trade Flows: Insights from Uganda. Bus. Econ. Res., 3(1), 227–239.
Mohamad, S., Jusoff, H.K. (2008). Exchange rates and export growth in Asian economies. Asian Soc. Sci., 4(11), 30–36. Mumbengegwi, C. (2002). Macroeconomic and Structural
Adjustment Policies In Zimbabwe. Palgrave Macmillan. Mzumara, M. (2012). An overview of Zimbabwe’s
macroeco-nomic environment. Int. J. Econ. Res., 3(1), 32–68. Nicita, A. (2013). Exchange Rates, International Trade and
Trade Policies. Int. Econ., 135–136(56), 47–61.
Nyanga, T., Sibanda, R., Kruger, J. (2020). Obstructions to Job Satisfaction: Voices of Small to Medium Enterprise Employees in the Manufacturing Industry in Zimba-bwe. Bus. Excell. Manag., 10(1), 16–30.
Obayelu, A.E., Salau, S. (2010). Agricultural Response to Prices and Exchange Rate in Nigeria: Application of Co-Integration and Vector Error Correction Model (VECM). J. Agric. Sci., 1(2), 73–81.
Odili, O. (2015). Real Exchange Rate Volatility, Economic Growth and International Trade in an Emerging Market Economy: Evidence from Nigeria. Int. J. Acad. Res. Bus. Soc. Sci., 5(7), 179–201.
Oiro, M.O. (2015). Real Exchange Rate Volatility and Exports in Kenya: 2005. J. World Econ. Res., 4(5), 115–31. Opoku-Afari, M. (2004). Measuring the Real Effective
Ex-change Rate (REER ) in Ghana. Nottingham. Retrieved from https://www.nottingham.ac.uk
Osman, M.A., Evans, S.R. (2005). Time Series Analysis of the Somalian Export Demand Equations: A Co-integration Approach. J. Econ. Soc. Res., 4(2), 71–92.
Pahlavani, M., Wilson, D., Worthington, A.C. (2005). Trade-GDP Nexus in Iran: An Application of the Autoregressive Distributed Lag (ARDL) Model School of Economics and Information Systems. Am. J. Appl. Sci., 2(7), 1158–1165. Payne, J.E., Mervar, A. (2010). Research note: The tourism– growth nexus in Croatia. Tour. Econ., 16(4), 1089–1094.
Reserve Bank of Zimbabwe (2014). Quarterly Economic
Re-view: Zimbabwe. Harare. Retrieved Oct 20th 2016 from:
www.rbz.co.zw/assets/quarterlyeconomic-review-decem-ber-2014-final.pdf
Reserve Bank of Zimbabwe (2016a). 2016 Mid-Term Mon-etary Policy Statement. Retrieved from: www.rbz.co.zw/ monetarypolicy.html Accessed 20 October 2016
Reserve Bank of Zimbabwe (2016b). Monetary Policy State-ment, Economic Transformation Through Transparency and Accountability. (January): 1–92. Retrieved Oct 20th 2016 from: www.rbz.co.zw/monetary-policy.html Saungweme, T. (2012). Trade Dynamics in Zimbabwe (1980–
2012). Rus. J. Agric. Soc.-Econ. Sci., 10(22), 31–34. Scoones, I., Marongwe, N., Mavedzenge, B., Murimbarimba,
F., Mahenehene, J., Sukume, C. (2011). Zimbabwe’s land reform: A summary of findings. Brighton, IDS.
Sithole, G. (1996). Analysis of Policy Reform and Structural Adjustment Programs in Zimbabwe With Emphasis on Agriculture and Trade. Washington, D.C.
Thaver, R.L., Bova, C. (2014). An Estimation of Ecuador’s Export Demand Function with the Us. Int. J. Bus. Fin. Res., 8(1), 89–102.
Tobacco Industry and Marketing Board (2015). 2015 Annual Statistical Report. Retrieved Oct 27th 2016 from: http:// www.timb.co.zw.106
Trade Economics (2020). Zimbabwe GDP per capita. Re-trieved April 30th 2020 from: https://tradingeconomics. com/zimbabwe/gdp-per-capita
United Nations Development Programme (2009). Foreign Trade, Competitiveness and the Balance of Payments. Ha-rare.
Wong, H.T. (2013). Real exchange rate misalignment and eco-nomic growth in Malaysia. J. Econ. Stud., 40(3), 298–313. World Bank (2016). Zimbabwe Economic Update, Harare.
Retrieved Oct 27th 2016 from: www.worldbank.org
Yazici, M. (2006). Is the J-curve effect observable in Turkish agricultural sector? J. Cent. Eur. Agric., 7(2), 319–322.
